Supplemental material to this article can be found at: http://dmd.aspetjournals.org/content/suppl/2018/07/06/dmd.118.082164.DC1

1521-009X/46/9/1372–1381$35.00 https://doi.org/10.1124/dmd.118.082164 DRUG METABOLISM AND DISPOSITION Drug Metab Dispos 46:1372–1381, September 2018 Copyright ª 2018 by The American Society for Pharmacology and Experimental Therapeutics Single Nucleotide Polymorphisms (SNPs) Distant from Xenobiotic Response Elements Can Modulate Aryl Hydrocarbon Receptor Function: SNP-Dependent CYP1A1 Induction s

Duan Liu, Sisi Qin, Balmiki Ray,1 Krishna R. Kalari, Liewei Wang, and Richard M. Weinshilboum

Division of Clinical Pharmacology, Department of Molecular Pharmacology and Experimental Therapeutics (D.L., S.Q., B.R., L.W., R.M.W.) and Division of Biomedical Statistics and Informatics, Department of Health Sciences Research (K.R.K.), Mayo Clinic, Rochester, Minnesota

Received April 22, 2018; accepted June 28, 2018

ABSTRACT Downloaded from CYP1A1 expression can be upregulated by the ligand-activated aryl fashion. LCLs with the AA genotype displayed significantly higher hydrocarbon receptor (AHR). Based on prior observations with AHR-XRE binding and CYP1A1 mRNA expression after 3MC estrogen receptors and estrogen response elements, we tested treatment than did those with the GG genotype. Electrophoretic the hypothesis that single-nucleotide polymorphisms (SNPs) map- mobility shift assay (EMSA) showed that oligonucleotides with the ping hundreds of base pairs (bp) from xenobiotic response elements AA genotype displayed higher LCL nuclear extract binding after (XREs) might influence AHR binding and subsequent expres- 3MC treatment than did those with the GG genotype, and mass dmd.aspetjournals.org sion. Specifically, we analyzed DNA sequences 5 kb upstream and spectrometric analysis of EMSA -DNA complex bands identified downstream of the CYP1A1 gene for putative XREs. SNPs located three candidate , two of which were co-immunoprecipitated 6500 bp of these putative XREs were studied using a genomic data– with AHR. In conclusion, we have demonstrated that the rs2470893 rich human lymphoblastoid cell line (LCL) model system. CYP1A1 SNP, which maps 196 bp from a CYP1A1 promoter XRE, is asso- mRNA levels were determined after treatment with varying concen- ciated with variations in 3MC-dependent AHR binding and CYP1A1 trations of 3-methylcholanthrene (3MC). The rs2470893 (21694G>A) expression. Similar “distant SNP effects” on AHR binding to an XRE SNP, located 196 bp from an XRE in the CYP1A1 promoter, was asso- motif and subsequent might occur for additional ciated with 2-fold variation in AHR-XRE binding in a SNP-dependent AHR-regulated . at ASPET Journals on September 24, 2021

Introduction P450 1A1, an enzyme that metabolizes xenobiotics and environmen- The aryl hydrocarbon receptor (AHR) is a ligand-activated transcrip- tal toxins, many of which are AHR ligands (Denison and Nagy, 2003). tion factor that is best known for its role in response to environmental Several XRE motifs have been identified across the CYP1A1 gene toxins (Denison and Nagy, 2003; Murray et al., 2014). Once activated, (Denison et al., 1988; Lo and Matthews, 2012). AHR forms a heterodimer with the AHR nuclear transporter (ARNT) CYP1A1 expression can vary widely, which theoretically might that binds to specific DNA sequence motifs, xenobiotic response contribute to variation in risk after carcinogen exposure (e.g., elements (XREs), resulting in the regulation of gene transcription lung cancer risk in smokers) (Rannug et al., 1995; Stucker et al., 2000). (Mimura and Fujii-Kuriyama, 2003). Although a structurally diverse As a result, many studies have been performed searching for possible group of chemical compounds are AHR ligands, the best characterized associations of CYP1A1 single nucleotide polymorphisms (SNPs) with high-affinity ligands are hydrophobic agents, such as halogenated the induction of CYP1A1 expression and carcinogenesis risk; however, — aromatic hydrocarbons and nonhalogenated polycyclic aromatic the results have often been contradictory with some studies, especially hydrocarbons (Denison and Nagy, 2003). A prototypic gene that is those performed in East Asia, reporting a correlation of CYP1A1 genetic induced by AHR activation is CYP1A1, which encodes cytochrome polymorphisms with CYP1A1 induction and/or cancer risk (Kawajiri et al., 1990; Kiyohara et al., 1998), whereas others have failed to replicate those observations (Inoue et al., 2000; Anttila et al., 2001; Smith et al., 2001). Most of those studies focused on two common This work was supported by the National Institutes of Health National Institute CYP1A1 SNPs: rs4646903, also known as CYP1A1*2A, which maps 39 of General Medical Sciences (NIGMS) [Grants U19 GM61388, R01 GM28157, and of CYP1A1 (Spurr et al., 1987), and rs1048943, a nonsynonymous SNP R01 GM125633]. 1Current affiliation: Assurex Health Inc., Mason, Ohio. that results in an Ile462Val substitution without change in enzyme https://doi.org/10.1124/dmd.118.082164. activity (Zhang et al., 1996; Persson et al., 1997). Both of these SNPs s This article has supplemental material available at dmd.aspetjournals.org. have higher minor allele frequencies in East Asian populations than

ABBREVIATIONS: 3MC, 3-methylcholanthrene; AHR, aryl hydrocarbon receptor; ARNT, AHR nuclear transporter; ChIP, chromatin immunopre- cipitation; Co-IP, coimmunoprecipitation; CYP1A1, cytochrome P450 1A1; DMSO, dimethyl sulfoxide; EMSA, electrophoretic mobility shift assay; ERE, estrogen response element; FIMO, Find Individual Motif Occurrences; GWAS, genome-wide association study; KD, knockdown/knocked down; LCL, lymphoblastoid cell line; MS, mass spectrometry; SNP, single nucleotide polymorphism; qRT-PCR, quantitative reverse transcription polymerase chain reaction; V, variant; WT, wild-type; XRE, xenobiotic response element.

1372 SNPs Distant from XREs Modulate AHR Function 1373 in Europeans, which might explain the contradictory results of attempts These genotype and gene expression data were deposited in the National Center to associate CYP1A1 genetic polymorphisms with CYP1A1 induction for Biotechnology Information Gene Expression Omnibus (GEO accession: and/or cancer risk; however, exactly how those two SNPs might alter GSE23120) (Niu et al., 2010). This LCL model system was previously used CYP1A1 activity remains unclear. Except for one AHR SNP (Smart and successfully to preform many studies of SNP-dependent gene transcription (Ingle Daly, 2000) that was reported to influence CYP1A1 induction, no SNPs et al., 2013; Ho et al., 2016; Liu et al., 2017). The LCLs were cultured in RPMI 1640 media containing 15% fetal bovine serum. have been associated with the induction of expression of this gene. As a Undifferentiated HepaRG human hepatic cells (HPR101) were obtained from result, the molecular basis for individual variation in the inducibility of Biopredict (Rennes, France) and were cultured and differentiated into functional CYP1A1 is not completely understood. hepatocyte-like cells according to the manufacture’s protocol. HPR101 cells were Recently, during studies of estrogen receptor a, we observed that grown in William’s E media (Gibco, Grand Island, NY) supplemented with 1Â SNPs at a distance (i.e., hundreds of base pairs) from estrogen response GlutaMAX (Gibco) and 10% HepaRG growth medium supplement (Biopredict) elements (EREs) could strongly influence both ER binding to those for 10–14 days until confluent. The cells were then switched into HepaRG EREs and subsequent gene expression (Ingle et al., 2013, 2016; Ho et al., Differentiation media for 2 weeks. Differentiated HepaRG cells were used to 2016; Liu et al., 2017). To determine whether this type of phenomenon preform transfection studies. might also occur with other nuclear receptors, we set out in the Human SK-N-BE(2) neuroblastoma and U-87 MG glioblastoma cell lines present study to test the hypothesis that SNPs at a significant were obtained from the American Type Culture Collection (ATCC; Manassas, VA). Original stocks of the cells were stored in liquid nitrogen and had not been distance from XREs might modulate AHR-XRE binding and influ- passaged. Both these cell lines were cultured in Eagle’s minimum essential ence the transcription of AHR-regulated genes such as CYP1A1,a

medium with 10% fetal bovine serum. Downloaded from process that might, in part, explain individual variation in CYP1A1 3-Methylcholanthrene (3MC) Treatment. A “prototypic” AHR ligand, a inducibility. ER acts in a fashion very similar to AHR. Once 3MC(Sigma,St.Louis,MI),wasusedtotreatLCLs,HepaRG,andU87MG activatedbyestrogenbinding,ERa forms a homodimer and binds to cells to activate AHR signaling pathways. Control cells were treated with specific DNA sequence motifs in EREs, thus regulating gene dimethyl sulfoxide (DMSO), which had been used to dissolve the 3MC. The expression. For example, a previous genome-wide association study DMSO final concentration in the cell culture media was 0.1% (v:v) for all of (GWAS) for response to tamoxifen breast cancer prevention therapy the control treatments. For CYP1A1 mRNA quantification, LCLs were treated identified SNP signals in intron two of the ZNF423 gene (Ingle et al., with 3MC at a series of concentrations ranging from 1E206 to 10 mMfor dmd.aspetjournals.org 2013). Functional genomic studies demonstrated that the ZNF423 24 hours and were then harvested for total RNA extraction. For chromatin m rs9940645 SNP, which maps 200 bp away from an ERE cluster, immunoprecipitation (ChIP) assays, cells were incubated with 0.1 M3MC for 2 hours and were then fixed by exposure to 1% formaldehyde. For reporter could influence ERa-ERE binding and ZNF423 transcription (Ingle gene assays, cells were treated with 0.1 mM 3MC for 24 hours before harvest et al., 2013). Furthermore, subsequent studies identified calmodulin-like for use in subsequent experiments. a “ ” protein 3 as a co-regulator of ER that sensed genotypes for the mRNA Quantification. mRNA levels were determined by quantitative ZNF423 intron two SNP (Qin et al., 2017). reverse transcription polymerase chain reaction (qRT-PCR) using the Power

In the present study, we aimed to determine whether SNPs at SYBR Green RNA-to-CT 1-Step kit (Applied Biosystems Inc., Foster City, CA). at ASPET Journals on September 24, 2021 a distance from XREs might modulate AHR-XRE binding and Total RNA was extracted from cells with the Quick-RNA MiniPrep Plus kit influence downstream gene transcription in a fashion similar to the (Zymo Research, Irvine, CA). A total of 100 ng of total RNA was used in each effect of the rs9940645 SNP in ZNF423 on ERa binding to EREs reaction. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) mRNA expres- by studying SNPs and XREs in and near CYP1A1. As described sion was measured for use as an internal control. Specific primers for CYP1A1, subsequently, we identified a SNP, rs2470893, that maps 59-of AHR, GAPDH mRNA, as well as other genes, were purchased from Integrated DNA Technologies (Coralville, IA). Primer sequences for those genes are listed in CYP1A1 that can influence both AHR binding to an XRE located Supplemental Table 1. 196 bp distant from the SNP and subsequent CYP1A1 transcription ChIP Assay. ChIP assays were conducted for LCLs using the EpiTect ChIP induced by the AHR agonist, 3-methylcholanthrene (3MC) (Riddick OneDay Kit (QIAGEN, Germantown, MD). In brief, approximately 5 Â 106 cells et al., 1994). We also identified possible AHR interacting proteins were treated with 0.1 mM 3MC for 2 hours. The same number of cells that had that might help to explain the mechanism responsible for this been treated with DMSO was used as a control. After the cells were cross-linked process. and harvested, 400 ml of lysis buffer was added to lyse the cells. DNA was sheared by sonication into 200- to 1000-bp fragments, with an average size of 500 bp. Aliquots of 100 ml of sheared chromatin/DNA was used for immunoprecipitation Materials and Methods with an AHR antibody (cat no. 83200S; Cell Signaling, Danvers, MA) or IgG Identification of Putative XREs. A motif scanning tool, Find Individual (cat. no. 2729S; Cell Signaling). Two microliters of purified ChIP DNA was Motif Occurrences (FIMO) (http://meme-suite.org/tools/fimo), was applied to added to a final 10-ml qPCR reaction mixture for the quantification of putative identify sequences matched to known XRE motifs (Denison et al., 1988; XRE sequences. SYBR Green ROX qPCR Mastermix (QIAGEN, Hilden, Sogawa et al., 2004). DNA sequences of the two known XRE motifs were used Germany) was used to amplify ChIP DNA. The qPCR primers used to amplify as templates for putative XRE scanning, the XRE-I motif with a sequence of specific DNA fragments, including putative XREs and SNPs, are listed in CACGCNA (Denison et al., 1988), and the XRE-II motif, with a sequence of Supplemental Table 2. Functional XREs were defined based on greater than

“CATGN6C[T/A]TG” (Sogawa et al., 2004). Based on known transcrip- 2-fold enrichment compared with DMSO-treated cells after exposure to AHR tional activity sites for the CYP1A1 gene included in ENCODE (https:// antibody in 3MC-treated cells. genome.ucsc.edu/ENCODE), DNA sequences located 5 kb upstream and Transient Transfection. LCLs were transfected with DNA plasmids for downstream of the CYP1A1 gene, as well as across the gene, were analyzed reporter gene assay or with siRNA for gene knockdown (KD) by electroporation using FIMO to identify putative XREs. using the Amaxa Cell Line Nucleofector Solution V Kit (Lonza, Cologne, Cell Culture. The Human Variation Panel of lymphoblastoid cell lines (LCLs) Germany). Specifically, LCLs were maintained at 1 Â 106 cells/ml for 2 days was obtained from the Coriell Institute (Camden, NJ). The DNA from these before use. For each transfection, 1 Â 106 cells were mixed with 2 mg of plasmid 300 LCLs had been genotyped in the Coriell Institute using the Affymetrix DNA or 100 pmol of siRNA in 100 ml of Nucleofector Solution V and incubated Human SNP Array 6.0 (Affymetrix, Santa Clara, CA) and in our laboratory using in cuvettes for 5 minutes without air bubbles. Sample suspensions in cuvettes were Illumina HumanHap 550K and HumanExon 510S-Duo BeadChips (Illumina, San then “pulsed” on a Nucleofector II Device (Lonza) with program X-001. After Diego, CA) to obtain genome-wide SNP data. Imputation was then performed the addition to the cuvette of 500 ml of pre-equilibrated culture media, samples using 1000 Genomes data. In addition, we generated gene expression data for all were gently transferred to culture plates and incubated at 37C overnight. Cells of the LCLs by using Affymetrix U133 2.0 Plus GeneChip expression arrays. were then treated with 3MC as described already before use. SK-N-BE(2) 1374 Liu et al. cells were transfected in the same fashion as LCLs using the Nucleofector II sequence motifs. Fifteen putative XREs were identified, of which seven Device but with the X-005 program. HepaRG and U87 MG cells were transfected matched XRE-I and eight matched XRE-II sequences (see Supplemental with reporter gene constructs and/or siRNA by using the Lipofectamine Table 5). We failed to identify any SNP that either disrupted or created 3000 Transfection Reagent (ThermoFisher, Grand Island, NY). Twenty-four XRE sequences. Since our purpose was to determine the possible hours after transfection, cell culture media were replaced with fresh media function of SNPs at a distance from XREs, SNPs located 6500 bp of containing 0.1 mM 3MC for an additional 24 hours. The cells were then harvested these putative XREs were included in the studies described subsequently for the extraction of total RNA and for luciferase activity. 6 Reporter Gene Assay. To determine the effects of SNP on CYP1A1 (Supplemental Table 5). We chose to study SNPs that mapped 500 bp transcription, an approximately 2.7-kb DNA segment that included sequence of an XRE because those SNPs could be studied with the greatest upstream (59) and exon 1 of CYP1A1 were subcloned upstream of the luc2 gene accuracy regarding AHR-XRE binding by the application of ChIP- (encoding firefly luciferase) in the pGL4-Basic vector (Promega, Madison, WI). qPCR assay with the amplification of both the XRE and SNP sequences DNA sequences of the primers used for subcloning were 59-ttgcCTCGAG- (Ingle et al., 2013; Liu et al., 2017). Specifically, during ChIP assays, the TAATTGGACGGAGCAAAAGG-39 (forward primer with the XhoI restriction chromatin is often sheared into fragments 200–1000 bp long, with an enzyme site underlined) and 59-gcgAAGCTTAGAAAGGGCAAGCCAGAAGT-39 average size of 500 bp (Nelson et al., 2006; Dahl and Collas, 2008; Saleh (reverse primer with the HindIII restriction enzyme site underlined). Genomic et al., 2008; Carey et al., 2009). Chromatin fragments 400–500 bp long DNA from LCLs with known genotypes for the SNP of interest was used as a have proven to be most suitable for PCR-based ChIP assays since they template for PCR reactions designed to amplify 2.7-kb DNA fragments containing cover two to three (Dahl and Collas, 2008). This length of wild-type (WT) or variant (V) SNP alleles. After confirmation of the sequence by the chromatin fragments limited our SNP selection since only fragments

DNA sequencing, reporter gene constructs containing WT or V SNP genotypes Downloaded from were transfected into cells as described already. A pRL-CMV vector (Promega) with AHR bound to the XRE would be precipitated by the AHR that expresses renilla luciferase was cotransfected as an internal control. After 3MC antibody, and only SNP(s) within the same chromatin fragment would treatment, the cells were lysed, and luciferase activity was determined using the be detected. SNPs located farther from the XRE being studied would not Dual-Luciferase Reporter assay system (Promega). be precipitated. To determine possible associations of SNPs with EMSA and MS. To determine whether the DNA sequence containing the CYP1A1 induction, we used a human lymphoblastoid cell line (LCL) SNP of interest might bind to nuclear proteins, EMSA was performed with nuclear model system consisting of cells from 300 individuals of three different extracts from LCLs after treatment with 3MC or DMSO. DNA oligonucleotides ethnic groups, for which genome-wide SNP genotype and mRNA dmd.aspetjournals.org (Oligoes) 21 nucleotides long with the SNP in the middle were synthesized by expression data had been generated in our laboratory (Niu et al., 2010). Integrated DNA Technologies. The DNA sequences of oligonucleotides for the Specifically, LCLs with WT or homozygous variant genotypes for the SNP of interest are listed in Supplemental Table 3. A similar set of oligonucleotides was synthesized with a biotin label at the 59-end. Double-stranded DNA fragments SNPs of interest were treated with an AHR agonist (3MC), followed by were obtained by annealing “sense” and “antisense” oligonucleotides, and those determination of CYP1A1 mRNA levels. We excluded SNPs having a dsDNA fragments were incubated with LCL nuclear extracts by using the minor allele frequency ,0.10 because it would be practically difficult to LightShift EMSA Optimization & Control Kit (Thermo Scientific, Rockford, IL). study those SNPs because of the limited number of LCLs homozygous

Nuclear extracts from LCLs treated with 3MC were prepared by using the Nuclear for variant SNP genotypes. Four SNPs within the area studied met these at ASPET Journals on September 24, 2021 Extract Kit (Active Motif, Carlsbad, CA). After incubation, the dsDNA-protein criteria, including the CYP1A1 MspI variant, and three other SNPs that complexes were separated in parallel using two 5% Mini-PROTEAN TBE Precast mapped 59 of the CYP1A1 gene (Table 1). Gels (Bio-Rad, Hercules, CA). After separation, one gel was used to transfer SNP-Dependent CYP1A1 Induction. To determine whether any of m dsDNA fragments to Biodyne B precut modified nylon membranes, 0.45 m these four SNPs (Table 1) might be associated with variation in CYP1A1 (Thermo Scientific). Biotin-labeled dsDNA on the membranes was then induction, we chose 14 LCLs based on their genotypes for these four visualized using the Chemiluminescent Nucleic Acid Detection module (Thermo Scientific). The other gel was stained using the Pierce Silver Stain SNPs (Supplemental Table 6). Since a previous GWAS that we had kit (Thermo Scientific). EMSA bands were then excised from the gel and sent performed for symptom severity of major depressive disorder had shown for mass spectrometry (MS) analysis. Sample processing and MS assays were that the AHR rs17137566 SNP was an eQTL in European populations performed by the Taplin Biologic Mass Spectrometry Facility at Harvard (Liu et al., 2018), and a previous study had reported that the AHR Medical School. rs2066853 SNP appeared to influence CYP1A1 induction (Smart and Coimmunoprecipitation. LCLs with homozygous variant genotypes (V/V) Daly, 2000), we excluded LCLs that were homozygous variant genotype for the rs2470893 SNP (n = 3) were treated with 3MC for 24 hours. Similar for those two AHR SNPs to avoid significant differences in baseline 7 amount of each LCL, approximately 10 cells, were mixed for total protein AHR activities in the 14 LCLs that we chose to study. These 14 LCLs extraction using NETN buffer (100 mM NaCl, 20 mM Tris-Cl (pH 8.0), 0.5 mM were then treated for 24 hours with 3MC at concentrations that ranged EDTA, and 0.5% Nonidet P-40) with protease and phosphatase inhibitor cocktails from 1E-06 to 10 mM. CYP1A1 mRNA levels were then determined by (Roche, Indianapolis, IN). Coimmunoprecipitation (Co-IP) was then performed as described previously (Yu et al., 2018). Antibodies against AHR, RCC1, NAP1L4, qRT-PCR. We found that the induction of CYP1A1 mRNA differed by and DDX39B or control IgG were used to perform Co-IP and Western blot genotype for the rs2470893 SNP (Fig. 1). This SNP was located 1570 bp analysis. Reverse Co-IP was performed using antibodies directed against RCC1, from the 59-end of the CYP1A1 transcription-start site, 196 bp from a NAP1L4, and UAP56, respectively. AHR antibody was used for the Western blot putative XRE (Fig. 1A). CYP1A1 induction in LCLs with homozygous analysis. Information about the antibodies used in the Co-IP and Western blot variant genotypes (V/V) for the rs2470893 SNP was significantly higher analyses is listed in Supplemental Table 4. than in LCLs homozygous for the WT genotype (W/W) (Fig. 1B). Statistical Analysis. Data analyses were performed, and graphs were plotted Genotypes for the three other SNPs listed in Table 1, either individually using GraphPad Prism (GraphPad Software, La Jolla, CA). Statistical compar- or in combination, were not associated with differences in CYP1A1 ’ t isons were made using Student s test or one-way analysis of variance. induction after 3MC treatment (data not shown). To determine whether this rs2470893 SNP-dependent difference in CYP1A1 induction might Results be caused by differences in AHR binding to the nearby putative XRE Identification of Putative XREs and SNP Selection. To test the sequence, ChIP assays with anti-AHR antibody were performed using hypothesis that SNPs distant from XREs in the CYP1A1 gene might LCLs with W/W and V/V genotypes after 2 hours of 3MC treatment. A influence its induction, we first identified putative XREs near and across significant increase in enrichment of SNP/XRE DNA fragments was the gene by scanning DNA sequences across and 5 kb upstream and observed in LCLs with the V/V genotype but not in those with the W/W downstream of the CYP1A1 gene for sequences that match known XRE genotype (Fig. 1C). Except for the rs2470893 SNP, we did not observe a SNPs Distant from XREs Modulate AHR Function 1375

TABLE 1 Common single nucleotide polymorphisms (SNPs) that mapped 6500 bp of xenobiotic response elements (XREs) across or 5 kb upstream and downstream of the CYP1A1 gene

Minor Allele Frequency (1000 Genomes Data) Distance (bp) Distance (bp) SNP rsID Common Allele Minor Allele Putative XRE Sequences Motif Typea P Valueb XRE to Genec SNP to XRE EUR EAS AFR AMR SAS rs17861109 C A 0.107 0.427 0.371 0.419 0.338 CACGCCA XRE-I 8.78E205 22106 259 rs2470893 G A 0.285 0.000 0.016 0.138 0.050 CAGGGAAGGCCTTG XRE-II 8.38E205 21766 196 rs3826041 A C 0.154 0.538 0.570 0.470 0.363 CACGCGA XRE-I 4.39E-05 21045 33 rs4646903 (MspI)T C 0.107 0.430 0.235 0.411 0.340 CACGCCA XRE-I 8.78E-05 294 53

AFR, African, n = 1322; AMR, admixed American, n = 694; EAS, East Asian, n = 1008; EUR, European, n = 1006; SAS, South Asian, n = 978. Base on 1000 Genomes data, the rs2470893 SNP was not in LD with the other three SNPs in European populations (r2 , 0.090). The other three SNPs were in LD with each other in European populations (r2 $ 0.630). The rs17861109 and rs4646903 SNPs were tightly linked (r2 5 1.000). a The XRE-I sequence motif is CACGCNA, and the XRE-II motif is CATGN6C[T/A]TG as described in the Materials and Methods section. bP value is given by the program FIMO to show the significance of the matching between putative XRE and the XRE motifs. cNegative and positive values refer to the distance from the 59-or39- ends of gene, respectively.

significant SNP-dependent difference in enrichment for the other three transfected with the variant construct showed 1.5-fold higher CYP1A1 Downloaded from SNP/XRE fragments in ChIP assays (data not shown). transcription activity than did cells transfected with the WT construct We selected LCLs for these initial studies of possible SNP-dependent (Fig. 2B). The same constructs were then transfected into HepaRG cells, CYP1A1 induction because of the availability of genome-wide SNP a cell line of hepatocyte origin that retains many of the characteristics of genotype information for these LCLs, which made it possible for us to human hepatocytes, including expression of most of the cytochrome select an adequate number of LCLs with the SNPs of interest for study. P450 enzymes (Guillouzo et al., 2007; Kanebratt and Andersson, 2008).

To determine whether the SNP-dependent CYP1A1 induction that we The variant genotype construct once again showed significantly greater dmd.aspetjournals.org observed in LCLs might also exist in other cell lines, we created reporter CYP1A1 transcriptional activity than did the WT construct (Fig. 2C). gene constructs that included approximately 2.7 kb of the CYP1A1 Since we had previously reported that AHR has a role in the regulation of promoter with WT and variant genotypes for the rs2470893 SNP kynurenine pathway genes in cells of central nervous system origin (Liu (Fig. 2A). Sequences of the WT and variant reporter gene constructs et al., 2018), the reporter gene constructs were also transfected into a were confirmed by DNA sequencing, which showed that they differed neuroblastoma cell line, SK-N-BE(2), and a glioblastoma cell line, U-87 only in the single nucleotide that defined the rs2470893 SNP. These MG. A significant increase in transcriptional activity was observed in

reporter gene constructs were then transfected into different cell lines SK-N-BE(2) cells transfected with the variant construct compared with at ASPET Journals on September 24, 2021 to compare CYP1A1 transcriptional activities for WT and variant SNP that for the WT construct (Fig. 2D), whereas no significant differences genotypes in those cells. were observed in U-87 MG (Fig. 2E). These reporter gene assay results The reporter gene constructs were first transfected into LCLs, and suggested that rs2470893 SNP-dependent CYP1A1 induction can be those cell lines were then treated with 3MC for 24 hours. LCLs cell line–dependent.

Fig. 1. (A) Schematic depiction of the location of the rs2470893 SNP and the nearby XRE at the 59- end of the CYP1A1 gene. Annealing sites for the qPCR primers that were used in the ChIP assay are indicated. (B) After 24-hours of treatment with 3MC at a series of concentrations, LCLs homo- zygous for the rs2470893 V/V genotype (n =6) showed significantly higher CYP1A1 mRNA expression than did LCLs homozygous for the W/W genotype (n = 8). (C) ChIP assays performed with anti-AHR antibody showed a 3-fold increase in AHR binding to the XRE 196 bp away from the V/V genotype rs2470893 SNP compared with those homozygous for the W/W genotype (n =3). *P , 0.05; **P , 0.01; ***P , 0.001. Values shown are mean 6 S.E.M. 1376 Liu et al.

Fig. 2. Reporter gene assays for the CYP1A1 rs2470893 SNP. (A) Structure of the CYP1A1 promotor reporter gene construct within

the basic pGL4 plasmid (Promega). The rs2470893 SNP was Downloaded from included in the CYP1A1 promotor sequence that was subcloned upstream of the luciferase reporter gene, Luc2. The plasmids contained either WT or variant sequences for the SNP within the CYP1A1 promoter and were transfected into (B) LCLs, (C) HepaRG, (D) SK-NB-E(2), and (E) U-87 MG cells. Values shown for (B–E) are ratios of relative light units (RLUs) compared with the internal reference after 3MC or DMSO vehicle treatment. **P , 0.01, ns, not

significant. Values shown are mean 6 S.E.M. for three determina- dmd.aspetjournals.org tions. Data for individual assays are shown in (B–E). at ASPET Journals on September 24, 2021

Identification of Possible rs2470893 SNP-Binding Proteins. To nine proteins identified only in the variant sample, were selected for test potential mechanisms for the rs2470893 SNP-dependent CYP1A1 further study of their possible involvement in SNP-specific CYP1A1 induction, EMSA was performed to identify potential AHR coregulator(s) induction (Fig. 3C). that might also bind to the SNP region (Fig. 3A). Specifically, nuclear To identify which of the rs2470893 SNP-binding proteins might extract from LCLs treated with 3MC was incubated with 21 bp of influence CYP1A1 induction in an SNP-dependent fashion, we knocked DNA fragments with the SNP at middle of the sequence. DNA fragments down (KD) each of the 28 candidate genes encoding these proteins in an with variant SNP genotype showed stronger binding to nuclear proteins LCL with homozygous variant rs2470893 SNP genotype, followed by than did those with the WT genotype (Fig. 3B). We then analyzed the quantification of CYP1A1 mRNA expression after treatment with EMSA bands by MS to identify proteins bound to the DNA fragments. vehicle (DMSO) or 3MC. The effects of the KD of AHR were also “Blank” samples excised from the margins of gel were also submitted determined as a positive control. CYP1A1 mRNA levels after the KD of for MS analysis as controls. The proteins identified in each sample by AHR and the 28 candidate genes were then normalized to the negative MS are listed in Supplemental Table 7. Proteins present in “blank” samples control KD (neg. siRNA). After treatment with either DMSO or 3MC, and proteins that are commonly seen and, as a result, might represent CYP1A1 mRNA levels were dramatically decreased after the KD of nonspecific binding owing to their abundance in cells (Malovannaya AHR compared with the control KD (Fig. 3D). Meanwhile, KD of et al., 2010; Hodge et al., 2013) were not pursued. Since MS identifies several of the candidate genes also resulted in a decrease in CYP1A1 proteins based on the identity of peptide fragments, proteins homologous mRNA levels (Fig. 3D). To determine which gene KD might have an to those present in “blank” samples were also not included in subsequent effect similar to that of AHR KD, CYP1A1 mRNA expression levels studies. In addition, proteins encoded by pseudogenes and proteins after the KD of each gene by siRNA transfection were plotted with that are not known to be present in the cell nucleus were excluded results for treatment with the DMSO vehicle on the x-axis and results for from further study. As a result, a total of 28 candidate proteins, treatment with 3MC on the y-axis (Fig. 3E). Each dot in the plot shown including 19 proteins common in both WT and variant samples and graphically in Fig. 3E represents results for gene KD by siRNA SNPs Distant from XREs Modulate AHR Function 1377

Fig. 3. (A) Schematic depiction of the hypothesis that coregulators might bind to the SNP and the AHR/ARNT complex to regulate CYP1A1 transcription. (B) Oligonucleotides 21 bp long with the SNP in the middle were synthesized for EMSA assays, which showed that oligonucleotides with the variant genotype displayed greater LCL nuclear extract binding after 3MC treatment than did those with the WT genotype. The EMSA bands were sent for MS analysis

to identify proteins that bound to the SNP DNA sequences. (C) Downloaded from Venn diagram for the MS data summary. A total of 28 proteins, including 19 that were present in both WT and variant bands and nine specific for the variant band, were selected for subsequent study. (D) Comparison of CYP1A1 mRNA expression levels after KD of 28 candidate proteins that had been identified by MS, together with data for the KD of AHR, followed by DMSO or 3MC treatment; *P , 0.05; **P , 0.01; ***P , 0.001. Values shown 6

are mean S.E.M. for three determinations. (E) Correlations of ratios dmd.aspetjournals.org of CYP1A1 mRNA expression after DMSO or 3MC treatment indicated that KD of the RCC1, NAP1L4,andDDX39B genes (see red arrows) resulted in CYP1A1 ratios similar to those found after the KD of AHR. Dots represent mean values for three determinations. at ASPET Journals on September 24, 2021

transfection, with its position in this figure dependent on CYP1A1 treatments that consisted of DMSO vehicle shown on the x-axis and mRNA level after treatment with either DMSO and 3MC. The dot for 3MC treatment results shown on the y-axis, dots for KD of those three AHR KD (siAHR) was closest to the origin (Fig. 3E), as anticipated, genes were once again located near the dot showing results for AHR KD because CYP1A1 mRNA expression was dramatically decreased after (Fig. 4A, right); however, in LCLs homozygous for the W/W rs2470893 the KD of AHR after treatment with the vehicle, DMSO, and 3MC. SNP genotype, KD of those three proteins did not change CYP1A1 CYP1A1 mRNA levels after the KD of Assembly Protein mRNA levels induced by 3MC treatment (Fig. 4B, left). In that case, 1 Like 4 (NAP1L4), Regulator of Condensation 1 (RCC1), dots depicting the results for KD of those three genes were close to that or DExD-Box Helicase 39B (DDX39B) were close to the position of for the nontargeting control KD (Fig. 4B, right), indicating that the results seen after AHR KD (Fig. 3E), raising the possibility that those NAP1L4, RCC1, and DDX39B proteins did not appear to signifi- three proteins might be involved in the AHR/ARNT transcriptional cantly affect CYP1A1 induction in LCLs with W/W genotypes for complex which binds near the rs2470893 SNP. the rs2470893 SNP. NAP1L4, RCC1, and DDX39B Contribute to rs2470893 SNP- Using reporter gene constructs for WT and variant SNP genotypes for Dependent CYP1A1 Induction. To determine whether NAP1L4, the rs2470893 SNP, we had demonstrated SNP-dependent CYP1A1 RCC1, and DDX39B proteins might contribute to SNP-dependent transcriptional activity in HepaRG cells (Fig. 2B). To determine whether CYP1A1 induction, the KD of those three genes was repeated in NAP1L4, RCC1, and DDX39B might influence CYP1A1 induction in LCLs with homozygous variant (V/V)andWT(W/W) rs2470893 HepaRG cells, these reporter gene assays were performed once again SNP genotypes, followed by CYP1A1 mRNA quantification. Although after separate KD of each of those three genes. In the control KD group CYP1A1 mRNA expression was significantly increased after treatment (neg. siRNA), CYP1A1 transcriptional activity was induced by 3MC with 3MC compared with DMSO vehicle treatment, KD of those three treatment, and the activity was once again significantly increased in cells genes, as well as the KD of AHR, significantly decreased CYP1A1 transfected with the variant reporter gene construct compared with induction compared with control KD in LCLs with a V/V genotype the results for the WT construct (Supplemental Fig. 1A). As expected, (Fig. 4A, left). After plotting CYP1A1 mRNA levels after KD of those KD of AHR (siAHR) repressed CYP1A1 transcriptional activity for three candidate genes in LCLs homozygous for V/V SNP genotypes with both WT and variant reporter gene constructs, and no significant 1378 Liu et al.

Fig. 4. (A, left) CYP1A1 induction by 3MC was decreased in LCLs with the rs2470893 V/V genotype after the KD of RCC1, NAP1L4, or DDX39B. (A, right) Correlation of CYP1A1 mRNA levels after DMSO or 3MC treatment showing that CYP1A1 mRNA expression after the KD of RCC1, NAP1L4, or DDX39B was similar to results obtained after the KD of AHR in rs2470893 V/V LCLs. (B, left) CYP1A1 induction by 3MC was unchanged in LCLs with the Downloaded from rs2470893 W/W genotype after the KD of RCC1, NAP1L4, or DDX39B, and (B, right), in contrast, CYP1A1 mRNA expression after the KD of RCC1, NAP1L4, or DDX39B was similar to data for the KD control in LCLs with the rs2470893 W/W genotype. *P , 0.05; **P , 0.01; ***P , 0.001. Values shown on the left are mean 6 S.E.M. for four determinations. Dots and squares shown on the right represent mean values for four determinations. dmd.aspetjournals.org at ASPET Journals on September 24, 2021

difference in 3MC-induced CYP1A1 transcriptional activity was found precipitation with anti-DDX39B antibody (data not shown). These between WT and V reporter gene constructs (Supplemental Fig. 1A). Co-IP studies confirmed that RCC1 and NAP1L4 could physically Furthermore, compared with control KD, KD of NAP1L4, RCC1, or interact with AHR. When joined with the EMSA assay results, CYP1A1 DDX39B in HepaRG cells also significantly repressed CYP1A1 mRNA expression results and the luciferase reporter gene assay transcriptional activity; as in the case of AHR KD, no differences in results, these observations suggest that RCC1 and NAP1L4 proteins CYP1A1 transcriptional activity were observed between WT and variant might interact with the rs2470893 SNP and enhance the binding of reporter gene constructs (Supplemental Fig. 1A). Finally, AHR, RCC1, AHR to the nearby XRE, thus contributing to the regulation of NAP1L4, and DDX39B mRNA levels were decreased significantly after CYP1A1 expression. specific siRNA transfections, as measured by qRT-PCR (Supplemental Fig. 1B). These results also indicated that NAP1L4, RCC1, and DDX39B might play a role in the SNP-dependent induction of CYP1A1 in Discussion HepaRG cells. Based on our previous observations with respect to the modulation RCC1 and NAP1L4 Interact Physically with AHR. To determine of ERa-ERE binding and downstream gene transcription (Ingle et al., whether NAP1L4, RCC1, or DDX39B might physically interact with 2013, 2016; Ho et al., 2016; Liu et al., 2017; Qin et al., 2017), in the the AHR transcriptional complex, coimmunoprecipitation (Co-IP) with present study, we set out to test the hypothesis that SNPs hundreds anti-AHR antibody was performed using LCLs treated with 3MC. After of bp distant from XREs might influence AHR-XRE binding and precipitation with anti-AHR antibody, the samples were analyzed by downstream gene expression. We chose to study CYP1A1, a pro- Western blot analysis to determine whether NAP1L4, RCC1 or totypical AHR-regulated gene, by analyzing DNA sequences across and DDX39B might precipitate with AHR. As shown in Fig. 5A, RCC1 and both up and downstream of the open reading frame for putative XREs NAP1L4 proteins were precipitated by the anti-AHR antibody, but and common SNPs that mapped 6 500 bp of those XREs. We identified DDX39B protein was not. To further confirm protein-protein interac- one SNP, rs2470893, located 59 of CYP1A1, that was associated with tions with AHR, “reverse” co-IP was performed using antibodies against significant variation in AHR binding to a nearby (196 bp distant) XRE NAP1L4, RCC1 and DDX39B, respectively, followed by detection and with CYP1A1 transcription in LCLs (Fig. 1). This rs2470893 SNP performed by AHR precipitation. AHR was detected after precipitation not only influenced CYP1A1 transcriptional activity in LCLs but with anti-RCC1 (Fig. 5B) and NAP1L4 (Fig. 5C) antibodies but not after also appeared to be functionally significant in cell lines of different SNPs Distant from XREs Modulate AHR Function 1379

the CYP1A1 rs2470893 SNP variant genotype results in enhanced AHR binding to a nearby XRE, with increased CYP1A1 transcrip- tion after 3MC treatment. Our results fit with an observation reported previously that the rs2470893 SNP was significantly associated with circulating polychlorinated biphenyl 118 levels in a general elderly population (n = 1016), with the variant genotype associated with a decrease in PCB118 concentrations (Lind et al., 2014). PCBs are AHR agonists and are metabolized by CYP1A1. The decrease in PCB118 plasma concentrations in subjects who carry the rs2470893 variant genotype might be a consequence of enhanced induction of CYP1A1 expression. The CYP1A1 and CYP1A2 genes are located “head to head”;asa result, they share the same 59 region on chromosome 15. The rs2470893 SNP is located between the two genes, approximately 1.5 kb upstream of CYP1A1 and 29.5 kb upstream of the CYP1A2 transcription start site. CYP1A2 is the major enzyme that metabolizes caffeine (Butler et al., 1989; Berthou et al., 1991). Several large-cohort GWAS studies for Downloaded from coffee consumption have identified rs2470893 as one of the “top hits” for association with that phenotype, with the variant genotype associated with less coffee consumption (Cornelis et al., 2011; Amin et al., 2012) and decreased caffeine metabolism (Cornelis et al., 2016). Caffeine was not found to induce CYP1A2 activity in human hepatocytes (Vaynshteyn and Jeong, 2012), but it did repress CYP1A1 expression in LCLs (Amin et al., 2012). It would be interesting to determine whether dmd.aspetjournals.org the rs2470893 variant genotype might enhance the caffeine effect on CYP1A2 repression and decrease caffeine metabolism; however, since CYP1A2 is not expressed in LCLs (data not shown), we were not able to study the possible effects of the SNP on CYP1A2 transcription in our cell line–based model system. Future studies using hepatocytes from individuals with different SNP genotypes might help to clarify Fig. 5. (A) Co-immunoprecipitation (Co-IP) with anti-AHR antibody identified interactions between AHR and RCC1 and NAP1L4 in LCLs that were treated with the possible role of this SNP, or other genetically linked SNPs, in at ASPET Journals on September 24, 2021 3MC but not for DDX39B. Reverse Co-IP with (B) anti-RCC1 and (C) anti- caffeine’s effect on CYP1A2. NAP1L4 antibodies confirmed interactions between AHR and RCC1 and between Functional genomic studies have largely been focused on genetic AHR and NAP1L4. The data shown are representative of the results of two individual experiments. variation that maps to exons that alter the encoded protein amino acid sequence or within canonical splice sequences that alter RNA splicing; however, most GWAS top-hit SNP signals map to introns tissue origins, including the hepatocyte-like HepaRG cells (Fig. 2). or intergenic regions, which indicates that noncoding SNPs play We then performed a series of studies to investigate possible molecular an important functional role, a role that often involves variation mechanisms for distant-SNP modulated AHR-XRE binding and gene in transcription regulation (Battle et al., 2017; Weinshilboum and transcription. Three proteins, NAP1L4, RCC1, and DDX39B were Wang, 2017; Wang et al., 2018). Gene transcription is often initiated identifiedbyMSafterEMSAassaythatcouldinteractwithDNA by transcription factor binding to specific DNA sequence motifs, sequences that included the rs2470893 SNP and which were with the AHR/ARNT heterodimer binding to XREs, followed by associated with AHR-dependent and SNP-dependent CYP1A1 recruitment of coregulators and RNA polymerase to form a transcrip- induction. Two of those proteins also interacted physically with tional complex (Malovannaya et al., 2011). We have previously AHR, which indicates that they may be involved in the mechanism reported that proteins encoded by members of the testis specific underlying distant-SNP modulated CYP1A1 transcription. Those Y-like (TSPYL) gene family were coregulators for the transcription findings provide, to our knowledge, the first example of SNPs that of cytochrome P450s, including CYP3A4, CYP2C9, CYP2C19, are near, but not in, an XRE and can regulate AHR-XRE binding and CYP17A1, and CYP19A1 (Liu et al., 2013; Qin et al., 2018). downstream gene transcription. Coregulators like members of the TSPYL gene family may, among Because of its important role in the metabolism of environmental other mechanisms, bind to specific DNA sequences, alter the toxins and procarcinogens, CYP1A1 genetic polymorphisms have nucleosome structure or recruit RNA polymerase, thus facilitating been studied for their possible association with the risk of the gene transcription (Qin et al., 2017). Therefore, changes in the occurrence of cancer, but the results have often been contradictory. binding of coregulators to DNA sequences as a result of SNPs that Those studies mainly focused on two common CYP1A1 SNPs, map hundreds of base pairs from response elements might well rs4646903 (also known as MspI mutant or CYP1A1*2A)andthe affect downstream gene transcription. In the present study, we rs1048943 (Ile462Val) SNP; however, the molecular mechanism by identified NAP1L4 and RCC1 as possible AHR coregulators that which those two SNPs might function in the regulation of CYP1A1 interact with rs2470893 SNP sequences and influence CYP1A1 activity remains unclear, which has hampered our understanding of transcription in LCLs and in HepaRG cells. NAP1L4, also known the possible relationship between CYP1A1 genetic polymorphisms as nucleosome assembly protein-2, is a chaperone protein and clinical phenotypes. By systematically testing a novel mech- that can transfer onto naked DNA templates (Rodriguez anism by which SNPs at a distance from XREs might modulate et al., 1997) and it functions within complexes with other proteins AHR function and CYP1A1 induction, we have demonstrated that that are distinct from histones (Rodriguez et al., 2004). RCC1 is a 1380 Liu et al. chromatin-bound protein that binds directly with the nucleosome Ho MF, Bongartz T, Liu M, Kalari KR, Goss PE, Shepherd LE, Goetz MP, Kubo M, Ingle JN, Wang L, et al. (2016) Estrogen, SNP-dependent chemokine expression and selective estrogen (Makde et al., 2010). Whether those two proteins regulate AHR receptor modulator regulation. Mol Endocrinol 30:382–398. transcriptional activity though interaction with nucleosomes needs Hodge K, Have ST, Hutton L, and Lamond AI (2013) Cleaning up the masses: exclusion lists to reduce contamination with HPLC-MS/MS. JProteomics88:92–103. to be determined in future studies. Finally, the DDX39B protein, Ingle JN, Liu M, Wickerham DL, Schaid DJ, Wang L, Mushiroda T, Kubo M, Costantino JP, which was shown to bind the rs2470893 SNP sequence and affect Vogel VG, Paik S, et al. (2013) Selective estrogen receptor modulators and pharmacogenomic variation in ZNF423 regulation of BRCA1 expression: individualized breast cancer prevention. CYP1A1 transcriptional activity (Fig. 3, D and E), did not physically Cancer Discov 3:812–825. interact with AHR (Fig. 5A). Whether DDX39B might interact with Ingle JN, Xie F, Ellis MJ, Goss PE, Shepherd LE, Chapman JW, Chen BE, Kubo M, Furukawa Y, other proteins recruited to the AHR-ARNT transcription complex Momozawa Y, et al. (2016) Genetic polymorphisms in the long noncoding RNA MIR2052HG offer a pharmacogenomic basis for the response of breast cancer patients to aromatase inhibitor needs to be determined. therapy. Cancer Res 76:7012–7023. In summary, by testing the hypothesis that SNPs distant from XRE Inoue H, Kiyohara C, Marugame T, Shinomiya S, Tsuji E, Handa K, Hayabuchi H, Onuma K, Hamada H, Koga H, et al. (2000) Cigarette smoking, CYP1A1 MspI and GSTM1 genotypes, motifs can modulate AHR transcriptional activity, we have demon- and colorectal adenomas. Cancer Res 60:3749–3752. strated that the rs2470893 SNP modulates AHR-dependent CYP1A1 Kanebratt KP and Andersson TB (2008) Evaluation of HepaRG cells as an in vitro model for “ ” human drug metabolism studies. Drug Metab Dispos 36:1444–1452. transcription. This distant SNP modulation of AHR transcriptional Kawajiri K, Nakachi K, Imai K, Yoshii A, Shinoda N, and Watanabe J (1990) Identification of activity might be a mechanism that, in part, explains variation in the genetically high risk individuals to lung cancer by DNA polymorphisms of the cytochrome P450IA1 gene. FEBS Lett 263:131–133. induction of CYP1A1 and, perhaps, other SNPs that influence the Kiyohara C, Nakanishi Y, Inutsuka S, Takayama K, Hara N, Motohiro A, Tanaka K, Kono S, expression of other AHR-regulated genes. RNA sequencing and ChIP and Hirohata T (1998) The relationship between CYP1A1 aryl hydrocarbon hy- – droxylase activity and lung cancer in a Japanese population. Pharmacogenetics 8: sequencing using LCLs and other genome data rich cell lines might – 315 323. Downloaded from result in the discovery of additional SNPs distant from XREs that Lind L, Penell J, Syvänen AC, Axelsson T, Ingelsson E, Morris AP, Lindgren C, Salihovic S, van modulate AHR-dependent gene transcription. Bavel B, and Lind PM (2014) Genetic variation in the CYP1A1 gene is related to circulating PCB118 levels in a population-based sample. Environ Res 133:135–140. Liu D, Ho MF, Schaid DJ, Scherer SE, Kalari K, Liu M, Biernacka J, Yee V, Evans J, Carlson E, et al. (2017) Breast cancer chemoprevention pharmacogenomics: deep sequencing and functional Authorship Contributions genomics of the ZNF423 and CTSO genes. NPJ Breast Cancer 3:30. Participated in research design: Liu, Qin, Ray, Weinshilboum. Liu D, Ray B, Neavin DR, Zhang J, Athreya AP, Biernacka JM, Bobo WV, Hall-Flavin DK, Conducted experiments: Liu, Qin, Kalari. Skime MK, Zhu H, et al. (2018) Beta-defensin 1, aryl hydrocarbon receptor and

plasma kynurenine in major depressive disorder: metabolomics-informed genomics. Transl dmd.aspetjournals.org Contributed new reagents and analytical procedures: Liu, Qin, Wang, Psychiatry 8:10. Weinshilboum. Liu M, Ingle JN, Fridley BL, Buzdar AU, Robson ME, Kubo M, Wang L, Batzler A, Jenkins GD, Performed data analysis: Liu, Qin, Kalari, Weinshilboum. Pietrzak TL, et al. (2013) TSPYL5 SNPs: association with plasma estradiol concentrations and aromatase expression. Mol Endocrinol 27:657–670. Wrote or contributed to the writing of the manuscript: Liu, Weinshilboum. Lo R and Matthews J (2012) High-resolution genome-wide mapping of AHR and ARNT binding sites by ChIP-Seq. Toxicol Sci 130:349–361. Makde RD, England JR, Yennawar HP, and Tan S (2010) Structure of RCC1 chromatin factor References bound to the nucleosome core particle. Nature 467:562–566. Amin N, Byrne E, Johnson J, Chenevix-Trench G, Walter S, Nolte IM, Vink JM, Rawal R, Malovannaya A, Lanz RB, Jung SY, Bulynko Y, Le NT, Chan DW, Ding C, Shi Y, Yucer N, Mangino M, Teumer A, et al.; kConFab Investigators (2012) Genome-wide association analysis Krenciute G, et al. (2011) Analysis of the human endogenous coregulator complexome. Cell – at ASPET Journals on September 24, 2021 Mol Psychiatry 145:787 799. of coffee drinking suggests association with CYP1A1/CYP1A2 and NRCAM. ’ 17:1116–1129. Malovannaya A, Li Y, Bulynko Y, Jung SY, Wang Y, Lanz RB, O Malley BW, and Qin J (2010) Anttila S, Tuominen P, Hirvonen A, Nurminen M, Karjalainen A, Hankinson O, and Elovaara E Streamlined analysis schema for high-throughput identification of endogenous protein com- – (2001) CYP1A1 levels in lung tissue of tobacco smokers and polymorphisms of CYP1A1 and plexes. Proc Natl Acad Sci USA 107:2431 2436. aromatic hydrocarbon receptor. Pharmacogenetics 11:501–509. Mimura J and Fujii-Kuriyama Y (2003) Functional role of AhR in the expression of toxic effects by – Battle A, Brown CD, Engelhardt BE, and Montgomery SB; GTEx Consortium; Labora- TCDD. Biochim Biophys Acta 1619:263 268. tory, Data Analysis &Coordinating Center (LDACC)—Analysis Working Group; Sta- Murray IA, Patterson AD, and Perdew GH (2014) Aryl hydrocarbon receptor ligands in cancer: – tistical Methods groups—Analysis Working Group; Enhancing GTEx (eGTEx) groups; friend and foe. Nat Rev Cancer 14:801 814. NIH Common Fund; NIH/NCI; NIH/NHGRI; NIH/NIMH; NIH/NIDA; Biospecimen Nelson JD, Denisenko O, and Bomsztyk K (2006) Protocol for the fast chromatin immunopre- – Collection Source Site—NDRI; Biospecimen Collection Source Site—RPCI; Bio- cipitation (ChIP) method. Nat Protoc 1:179 185. specimen Core Resource—VARI; Brain Bank Repository—University of Miami Brain Niu N, Qin Y, Fridley BL, Hou J, Kalari KR, Zhu M, Wu TY, Jenkins GD, Batzler A, Endowment Bank; Leidos Biomedical—Project Management; ELSI Study; Genome and Wang L (2010) Radiation pharmacogenomics: a genome-wide association approach to Browser Data Integration &Visualization—EBI; Genome Browser Data Integration identify radiation response biomarkers using human lymphoblastoid cell lines. Genome Res – &Visualization—UCSC Genomics Institute, University of California Santa Cruz; Lead 20:1482 1492. analysts; Laboratory, Data Analysis &Coordinating Center (LDACC); NIH program Persson I, Johansson I, and Ingelman-Sundberg M (1997) In vitro kinetics of two human CYP1A1 management; Biospecimen collection; Pathology; eQTL manuscript working group variant enzymes suggested to be associated with interindividual differences in cancer suscepti- (2017) Genetic effects on gene expression across human tissues. Nature 550:204–213. bility. Biochem Biophys Res Commun 231:227–230. Berthou F, Flinois JP, Ratanasavanh D, Beaune P, Riche C, and Guillouzo A (1991) Evidence for QinS,IngleJN,LiuM,YuJ,WickerhamDL,KuboM,WeinshilboumRM,andWangL the involvement of several cytochromes P-450 in the first steps of caffeine metabolism by human (2017) Calmodulin-like protein 3 is an estrogen receptor alpha coregulator for gene ex- liver microsomes. Drug Metab Dispos 19:561–567. pression and drug response in a SNP, estrogen, and SERM-dependent fashion. Breast Butler MA, Iwasaki M, Guengerich FP, and Kadlubar FF (1989) Human cytochrome P-450PA Cancer Res 19:95. (P-450IA2), the phenacetin O-deethylase, is primarily responsible for the hepatic Qin S, Liu D, Kohli M, Wang L, Vedell PT, Hillman DW, Niu N, Yu J, Weinshilboum RM, 3-demethylation of caffeine and N-oxidation of carcinogenic arylamines. Proc Natl Acad Sci and Wang L (2018) TSPYL family regulates CYP17A1 and CYP3A4 expression: potential USA 86:7696–7700. mechanism contributing to abiraterone response in metastatic castration-resistant prostate cancer. Carey MF, Peterson CL, and Smale ST (2009) Chromatin immunoprecipitation (ChIP). Cold Clin Pharmacol Ther 104:201–210. Spring Harb Protoc 2009:pdb prot5279. Rannug A, Alexandrie AK, Persson I, and Ingelman-Sundberg M (1995) Genetic poly- Cornelis MC, Kacprowski T, Menni C, Gustafsson S, Pivin E, Adamski J, Artati A, Eap CB, morphism of cytochromes P450 1A1, 2D6 and 2E1: regulation and toxicological Ehret G, Friedrich N, et al.; Swiss Kidney Project on Genes in Hypertension (SKIPOGH) significance. J Occup Environ Med 37:25–36. team (2016) Genome-wide association study of caffeine metabolites provides new insights Riddick DS, Huang Y, Harper PA, and Okey AB (1994) 2,3,7,8-Tetrachlorodibenzo-p-dioxin to caffeine metabolism and dietary caffeine-consumption behavior. Hum Mol Genet 25: versus 3-methylcholanthrene: comparative studies of Ah receptor binding, transformation, and 5472–5482. induction of CYP1A1. J Biol Chem 269:12118–12128. Cornelis MC, Monda KL, Yu K, Paynter N, Azzato EM, Bennett SN, Berndt SI, Boerwinkle E, Rodriguez P, Munroe D, Prawitt D, Chu LL, Bric E, Kim J, Reid LH, Davies C, Naka- Chanock S, Chatterjee N, et al. (2011) Genome-wide meta-analysis identifies regions on 7p21 gama H, Loebbert R, et al. (1997) Functional characterization of human nucleosome (AHR) and 15q24 (CYP1A2) as determinants of habitual caffeine consumption. PLoS Genet 7: assembly protein-2 (NAP1L4) suggests a role as a histone chaperone. Genomics 44: e1002033. 253–265. Dahl JA and Collas P (2008) A rapid micro chromatin immunoprecipitation assay (microChIP). Nat Rodriguez P, Ruiz MT, Price GB, and Zannis-Hadjopoulos M (2004) NAP-2 is part of multi- Protoc 3:1032–1045. protein complexes in HeLa cells. J Cell Biochem 93 :398–408. Denison MS, Fisher JM, and Whitlock JP, Jr (1988) The DNA recognition site for the diox- Saleh A, Alvarez-Venegas R, and Avramova Z (2008) An efficient chromatin immunoprecipitation in-Ah receptor complex. Nucleotide sequence and functional analysis. JBiolChem263: (ChIP) protocol for studying histone modifications in Arabidopsis plants. Nat Protoc 3: 17221–17224. 1018–1025. Denison MS and Nagy SR (2003) Activation of the aryl hydrocarbon receptor by struc- Smart J and Daly AK (2000) Variation in induced CYP1A1 levels: relationship to CYP1A1, Ah turally diverse exogenous and endogenous chemicals. Annu Rev Pharmacol Toxicol 43: receptor and GSTM1 polymorphisms. Pharmacogenetics 10:11–24. 309–334. Smith GB, Harper PA, Wong JM, Lam MS, Reid KR, Petsikas D, and Massey TE (2001) Human Guillouzo A, Corlu A, Aninat C, Glaise D, Morel F, and Guguen-Guillouzo C (2007) The human lung microsomal cytochrome P4501A1 (CYP1A1) activities: impact of smoking status and hepatoma HepaRG cells: a highly differentiated model for studies of liver metabolism and CYP1A1, aryl hydrocarbon receptor, and glutathione S-transferase M1 genetic polymorphisms. toxicity of xenobiotics. Chem Biol Interact 168:66–73. Cancer Epidemiol Biomarkers Prev 10:839–853. SNPs Distant from XREs Modulate AHR Function 1381

Sogawa K, Numayama-Tsuruta K, Takahashi T, Matsushita N, Miura C, Nikawa J, Gotoh O, Weinshilboum RM and Wang L (2017) Pharmacogenomics: precision medicine and drug response. Kikuchi Y, and Fujii-Kuriyama Y (2004) A novel induction mechanism of the rat CYP1A2 Mayo Clin Proc 92:1711–1722. gene mediated by Ah receptor-Arnt heterodimer. Biochem Biophys Res Commun 318: Yu J, Qin B, Moyer AM, Nowsheen S, Liu T, Qin S, Zhuang Y, Liu D, Lu SW, Kalari KR, et al. 746–755. (2018) DNA methyltransferase expression in triple-negative breast cancer predicts sensitivity to Spurr NK, Gough AC, Stevenson K, and Wolf CR (1987) Msp-1 polymorphism detected decitabine. J Clin Invest 128:2376–2388. with a cDNA probe for the P-450 I family on chromosome 15. Nucleic Acids Res 15: Zhang ZY, Fasco MJ, Huang L, Guengerich FP, and Kaminsky LS (1996) Characterization of 5901. purified human recombinant cytochrome P4501A1-Ile462 and -Val462: assessment of a role for Stücker I, Jacquet M, de Waziers I, Cénée S, Beaune P, Kremers P, and Hémon D (2000) the rare allele in carcinogenesis. Cancer Res 56:3926–3933. Relation between inducibility of CYP1A1, GSTM1 and lung cancer in a French population. Pharmacogenetics 10:617–627. Vaynshteyn D and Jeong H (2012) Caffeine induces CYP1A2 expression in rat hepatocytes but not Address correspondence to: Dr. Richard M. Weinshilboum, Mayo Clinic, 200 First in human hepatocytes. Drug Metab Lett 6:116–119. Wang L, Ingle J, and Weinshilboum R (2018) Pharmacogenomic discovery to function and Street SW, Rochester, MN 55905. E-mail: [email protected] mechanism: breast cancer as a case study. Clin Pharmacol Ther 103:243–252. Downloaded from dmd.aspetjournals.org at ASPET Journals on September 24, 2021